By Chris Hart, Manager of Engineering & Analytics
This is the second part in a three-part series on Demand Charges. In the first installment, we covered the basics of demand and demand charge. This time we’ll address how solar can (in theory) be used to reduce the demand charge portion of your electric bill. We’ll also discuss the real-world challenges of demand charge reduction, as well as the challenges of accurately modeling these reductions.
Let’s begin by looking at a building’s load curve on a typical summer weekday, when typically its electricity load is at its highest. In most cases, a building load is driven primarily by occupants and the services they demand — HVAC, office equipment, and lighting. This means that high loads tend to occur during daylight hours. The figure below shows actual building data as well as the theoretical net system load if connected to a 300kW rooftop solar system (for reference, the peak demand of the building for the year is 250kW). The solar system production is modeled based on data from a National Renewable Energy Lab (NREL) database of sun hours. The red line shows the original demand, while the purple area shows the net energy consumption with solar and the green area is the energy saved from adding solar to the system. It needs to be noted that the NREL data uses a 30 year average meteorological day when calculating the power produced from the solar system and does not take into account difficult to predict weather events.
At first glance, it might appear that there should be a significant demand charge savings. After all, the original peak demand was almost completely mitigated. This is a classic demand charge pitfall. To truly understand what the savings will be, we must first look at the electric rate structure applicable for this property. As detailed in Part 1, properties on PG&E’s E-19 rate (Medium General Demand-Metered TOU Service) pay three separate demand charges: 1) maximum demand (at any point in the day), 2) maximum peak (between 12pm and 6pm) and 3) part peak (between 8am and 12pm or 6pm and 9pm). Based on the actual load data, this building would pay $7,154 for this month of demand charges.
Adding solar can often have non-obvious effects, such as shifting the time at which the peak demand occurs, but not actually reducing the realized demand charge by significant amounts. Take our example: the maximum peak before solar occurred at 3pm and was 222kW. With solar, the demand at 3pm is reduced to 87kW – a 61% reduction! However, at 6pm (still during peak hours) the demand is 179kW, which is only a 19% reduction. The next chart overlays the time of use hours to better illustrate this concept.
Another issue to consider is that not all demand charges are created equally; some are more expensive than others. For example, for the E-19 tariff during summer months, while the overall maximum demand is worth $12.56/kW, the peak maximum demand charge increases to $16.13/kW, and the part-peak max drops to $3.74/kW. If we make the assumption that the example day we’ve been evaluating is the day of the month with highest demand, we would see a demand charge savings of $1,056. This represents a savings of 15% on the original demand bill. For reference, this 300 kW system offsets 28% of the total kWh consumed that day, and 35% of the energy bill (based on TOU energy charges). The table below shows a summary of the demand charge savings.
Recall that these savings are calculated assuming an idealized, theoretical solar generation system which doesn’t include potential major changes to solar output that can occur in short amounts of time . NREL’s meteorological database is good for predicting energy generation, but not for demand mitigations because it ignores micro-weather events. For example, if an afternoon thunderstorm rolls through on the day which the building’s energy use is the highest for the month (e.g. hottest summer day of the month), the output from the solar system can drop to nearly zero and the building’s energy will have to be drawn entirely from the grid. If this sudden increase in demand from the utility lasts more than 15 minutes, this high point in demand will become the demand peak for the month. Remember also that demand charges are based on peak demands for an entire month, so all it takes is one ill-timed thunderstorm a month to ruin all potential demand charge savings.
The randomness of these events means that it can be very difficult to accurately model demand charge savings. The savings calculated above represent a best-case scenario (no weather events) that can be used as an upper bound on potential savings, but shouldn’t be taken as a realistic possibility. Often times, project developers will include this best-case demand charge reduction scenario to make marginal projects seem more attractive to clients. Without some sort of probabilistic discount (either via Monte Carlo or other statistical analysis), projected demand charge savings are disingenuous or simply misleading, at best, and should not be used in cash flows when determining project returns. Alta Energy’s experienced finance and engineering team can help you determine if proposals your company has received include unrealistic demand charge savings.
In the next post, we’ll conclude the demand charge topic with a discussion of some strategies and technologies currently being developed that can help deliver more reliable demand charge savings.